1256 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO.

7, SEPTEMBER 2009
Spectrum Sharing between Cellular and Mobile Ad
Hoc Networks: Transmission-Capacity Trade-Off
Kaibin Huang, Vincent K. N. Lau, and Yan Chen
AbstractSpectrum sharing between wireless networks im-
proves the efciency of spectrum usage, and thereby alleviates
spectrum scarcity due to growing demands for wireless broad-
band access. To improve the usual underutilization of the cellular
uplink spectrum, this paper addresses spectrum sharing between
a cellular uplink and a mobile ad hoc networks. These networks
access either all frequency subchannels or their disjoint subsets,
called spectrum underlay and spectrum overlay, respectively. Given
these spectrum sharing methods, the capacity trade-off between
the coexisting networks is analyzed based on the transmission
capacity of a network with Poisson distributed transmitters. This
metric is dened as the maximum density of transmitters subject
to an outage constraint for a given signal-to-interference ratio
(SIR). Using tools from stochastic geometry, the transmission-
capacity trade-off between the coexisting networks is analyzed,
where both spectrum overlay and underlay as well as successive
interference cancelation (SIC) are considered. In particular, for
small target outage probability, the transmission capacities of
the coexisting networks are proved to satisfy a linear equation,
whose coefcients depend on the spectrum sharing method and
whether SIC is applied. This linear equation shows that spectrum
overlay is more efcient than spectrum underlay. Furthermore,
this result also provides insight into the effects of network
parameters on transmission capacities, including link diversity
gains, transmission distances, and the base station density. In
particular, SIC is shown to increase the transmission capacities of
both coexisting networks by a linear factor, which depends on the
interference-power threshold for qualifying canceled interferers.
Index TermsSpatial reuse; wireless networks; Poisson pro-
cesses; spectrum sharing; interference cancelation
I. INTRODUCTION
D
ESPITE spectrum scarcity, most licensed spectrums are
underutilized according to the Federal Communications
Commission [1]. In particular, in existing cellular systems
based on frequency division duplex (FDD) such as FDD
UMTS [2], equal bandwidths are allocated for uplink and
downlink transmissions, even though the data trafc for down-
link is much heavier than that for uplink [3], [4]. Spectrum
sharing between wireless networks improves spectrum utiliza-
tion, and will be a key solution for broadband access in next-
generation wireless networks [5]. This motivates the study in
this paper on sharing the uplink spectrum between a cellular
network and a mobile ad hoc network (MANET), called the
Manuscript received 30 August 2008; revised 31 January 2009.
K. Huang is with the School of Electrical and Electronic Engineering,
Yonsei University, 262 Seongsanno, Seodaemun-gu Seoul 120-749, Korea (e-
mail: huangkb@yonsei.ac.kr).
V. K. N. Lau is with the Department of Electronic and Computer Engi-
neering, Hong Kong University of Science and Technology, Clear Water Bay,
Hong Kong (e-mail: eeknlau@ust.hk).
Y. Chen is with Huawei Technology, Pudong, Shanghai, China (e-mail:
yanchen@ust.hk).
Digital Object Identier 10.1109/JSAC.2009.090921.
coexisting networks. A basic question is what is the trade-off
between the capacities of these networks?
We provide answers to this question using the metric of
transmission capacity [6], [7]. By extending the denition
in [6], the transmission capacities of the coexisting networks
are dened as their maximum transmitter densities under an
outage probability constraint for a target signal-to-interference
ratio (SIR). We derive the transmission-capacity trade-off
between the networks for different spectrum-sharing methods.
Such results are useful for controlling the sizes of the coex-
isting networks for optimizing uplink spectrum usage.
A. Related Works and Motivation
A spectrum band can be either licensed or unlicensed, where
a licence gives a network the exclusive right of spectrum
usage. Depending on whether holding a licence, a wireless
network is referred to as the primary (e.g. cellular networks)
or secondary network (e.g. MANETs). Accessing a licensed
band, the secondary transmitters must not cause signicant
interference to the primary receivers. One simple method
of sharing a licensed band is to spread the signal energy
radiated by each secondary transmitter over the whole band
using spread spectrum techniques [8], suppressing the power
spectrum density of the resultant interference to the primary
receivers. This method is called spectrum underlay [1], [5],
[9], [10].
Another method for sharing a licensed spectrum is called
spectrum overlay, where secondary transmitters access fre-
quency subchannels unused by nearby primary receivers.
Recent research on spectrum overlay has been focusing on
designing cognitive-radio algorithms for secondary transmit-
ters to opportunistically access the spectrum [5], [9]. Such
algorithms require secondary transmitters to continuously de-
tect and track transmission opportunities by spectrum sensing,
and decide on transmission based on sensing results [11], [12].
These operations are vulnerable to sensing errors, and most
important require complicated computation at the secondary
transmitters. For this reason, we consider the case where base
stations in the cellular (primary) network coordinates spectrum
sharing. Thereby ad hoc (secondary) transmitters use a simple
random access protocol rather than complicated cognitive-
radio algorithms.
In unlicensed spectrums such as the industrial, scientic
and medical (ISM) bands, all networks have equal priorities
for spectrum access. The networks using unlicensed bands
include wireless local area networks (WLANs) and wireless
personal area networks (WPANs). Due to mutual interference,
the coexistence of networks in unlicensed bands degrades the
0733-8716/09/$25.00 c 2009 IEEE
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
HUANG et al.: SPECTRUM SHARING BETWEEN CELLULAR AND MOBILE AD HOC NETWORKS: TRANSMISSION-CAPACITY TRADEOFF 1257
networks performance as shown by analysis [13], [14], sim-
ulation [15], [16], and measurement [17]. Sharing unlicensed
bands between competing networks is also studied using game
theory [18].
There exist few theoretical results on the network capacity
trade-off between the coexisting networks despite this being
a fundamental issue. In [19], the transmission capacities of
a two-tier network are analyzed, which comprises a cellular
network and a network of femtocell hot-spots. In [20], the
transport capacities
1
of two coexisting multi-hop MANETs are
shown to follow the optimum scaling laws for asymptotically
large network sizes. In [19], [20], the network-capacity trade-
off between coexisting networks is not analyzed.
The metric of transmission capacity has been applied to de-
signing MANETs with Poisson distributed transmitters and an
ALOHA-like medium-access-control layer, addressing issues
including spatial diversity [22], opportunistic transmission [7],
bandwidth partitioning [23], successive interference cance-
lation (SIC) [24], and spatial interference cancelation [25].
Besides the transmission capacity, a new performance metric
called capacity region is used in this paper, which is tailored
for spectrum sharing. The capacity region of the coexisting
networks is dened as the set of feasible combinations of
transmitter densities under the SIR outage constraints. By
denition, transmission capacities of the coexisting networks
give the boundary points of the capacity region. Both the
transmission capacities and capacity region of the coexisting
networks are analyzed in this paper.
B. Contributions and Organization
This paper targets a cellular uplink network and a MANET
sharing the uplink spectrum using either spectrum overlay
or underlay, where uplink users, base stations, and ad hoc
transmitters all follow Poisson distributions but with different
densities. Each transmitter modulates signals using frequency-
hopping spread spectrum over the frequency subchannels
assigned to the corresponding network [8]. This modulation
decreases the density of transmitters accessing the same sub-
channel. As a result, the inter-link interference reduces and
the transmission capacities increase.
Our main contributions are summarized as follows. First,
considering an interference-limited environment, bounds on
the SIR outage probabilities are derived for spectrum over-
lay and underlay with and without using SIC at receivers
[24], [26]. Second, for small target outage probability, the
transmission-capacities of the coexisting networks are showed
to satisfy a linear equation, whose coefcients depend on the
spectrum-sharing method and whether SIC is used. Given the
above linear capacity trade-off, the capacity region of the co-
existing networks is a triangular. Third, for small target outage
probability, the capacity region for spectrum underlay is shown
to be no larger than that for spectrum overlay. The former can
be enlarged to be identical to the latter by choosing the derived
transmission-power ratio between the two networks. Finally,
we characterize the effects of different parameters on network
transmission capacities. In particular, for spectrum overlay,
1
This metric introduced in [21] refers to end-to-end throughput per unit
distance of a multi-hop wireless network.
the transmission capacity C of the cellular network increases
linearly with the base station density
b
, and the capacity

C
of the MANET grows inversely with the decreasing distance

d
between a pair of communicating ad hoc nodes. For spectrum
underlay, C and

C both increase linearly with
b
or otherwise
inversely with decreasing

d. Moreover, C and

C both grow
sub-linearly with spatial diversity gains. Furthermore, SIC
increases both C and

C by a linear factor that is a function
of the interference-power threshold for qualifying canceled
interferers.
To distinguish our contributions from exiting results, several
new challenges addressed in the current work are emphasized.
First, the current analysis of the network outage probabilities
requires characterizing the distribution of sum interference
power from two classes of interferers from separate networks,
which differ in properties such as transmission power and
densities. This issue does not exist in prior works focusing
on stand-alone networks. Second, due to spectrum sharing, the
transmission capacities of the coexisting networks must satisfy
simultaneously two SIR outage constraints rather than one
when a stand-alone network is considered. Consequently, the
current transmission-capacity analysis must account for new
issues including heterogeneous network architectures and sizes
as well as resource allocation cross networks. These issues
complicate the relation between the transmission capacities of
the coexisting networks. Whether a simple relation exists is
unknown from prior works. Finally, we generalize the study
of SIC in [24] for a MANET with path-loss to the coexisting
networks with both fading and path-loss.
The remainder of this paper is organized as follows. The
mathematical models are described in Section II. The back-
ground is provided in Section III. In Section IV, the bounds on
outage probabilities are derived for different spectrum sharing
methods. For small target outage probability, the transmission-
capacity trade-off is analyzed in Section V. Numerical results
are presented in Section VI, followed by concluding remarks
in Section VII.
II. MATHEMATICAL MODELS
A. Network Models
The coexisting networks are illustrated in Fig. 1. Their
models are described as follows.
Following [6], [7], [27], we make the following assumption.
Assumption 1. The transmitters in the MANET form a Pois-
son point process (PPP) on the two-dimensional plane, which
is denoted as

with the density

.
This Poisson assumption as well as the similar one in Assump-
tion 2 are made for mathematical tractability. However, they
are supported by experiments for networks with mobility. In
particular, network models based on similar assumptions have
been employed by cellular service providers to successfully
predict network trafc load and blocking probabilities [28].
Next, each transmitter in the MANET is associated with a
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
1258 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 7, SEPTEMBER 2009
Base station
Uplink user
Ad hoc node
Fig. 1. The coexisting cellular and ad hoc networks
receiver located at a xed distance

d.
2
The transmission power
of transmitters is assumed xed and denoted as

P.
The cellular network is modeled based on the following
assumption
Assumption 2. The base stations and uplink mobiles form
two independent stationary PPPs denoted as and , re-
spectively.
Their corresponding densities are represented by
b
and .
Let B
n
, U
m
, D
n,m
denote the two-dimensional coordinates
of the nth base station, the mth uplink mobile, and their
distance, respectively. Thus, D
n,m
= |B
n
U
m
|.
3
To en-
hance the long-term link reliability, each mobile transmits to
the nearest base station. Consequently, the cellular network
forms a Poisson tessellation of the two-dimensional plane
and each cell is known as a Voronoi cell [29]. The mo-
biles served by the mth base station form the set V
m
:=
{U ||U B
m
| < |U B| B \{B
m
}}.
A typical point of a PPP is dened as a point selected
using the procedure where every point of the PPP has the
same probability of being selected [29]. The typical points
of the PPPs , and

are referred to as the typical base
station B
0
, the typical mobile U
0
, and the typical ad hoc
transmitter T
0
, respectively. Moreover, the intended receiver
for T
0
is called the typical ad hoc receiver and denoted as R
0
.
Finally, represent the random distance between B
0
and U
0
as
D := |B
0
U
0
|.
B. Channel and Modulation
The uplink spectrum is divided into M frequency-at sub-
channels by using orthogonal frequency division multiplexing
(OFDM) [30]. Each of the coexisting networks uses a subset
or the full set of subchannels, depending on the spectrum
2
Consideration of the randomness in

d provides little new insight. It is
straightforward to extend the results in this paper to include the randomness
in

d. Specically, if

d is random, Proposition 1-4 still apply except that the
expectation operations in Proposition 1-2 must account for the randomness
of

d. Furthermore,

d in Theorem 1 should be replaced with E[

d].
3
The operator |X| gives the Euclidean distance between X and the origin
if X is two-dimensional coordinates, or the cardinality of X if X is a set.
sharing methods discussed in Section II-C. In each network, a
transmitter modulates signals using frequency-hopping spread
spectrum, where signals hop randomly over all subchannels
assigned to the afliated network [6], [8].
Consider the link between U
0
and B
0
. The typical subchan-
nel accessed by U
0
and B
0
comprises path loss and a fading
factor denoted by W such that the signal power received by B
0
is PWD

, where P is the transmission power. Similarly, the

interference power from an interferer X to B
0
is P
X
G
X
R

X
,
where P
X
{P,

P}, and R
X
= |X B
0
|, and G
X
is the
fading factor.
Similar channel models are used for the MANET. Speci-
cally, the received signal power for T
0
is

P

W

d

where

W
is the fading factor; the interference power from an interferer
X to T
0
is P
X
G
X

R

X
where

R
X
= |X T
0
|.
C. Spectrum Sharing Methods
1) Spectrum Overlay: For spectrum overlay, the M sub-
channels are divided into two disjoint subsets and assigned to
two coexisting networks. Let A and

A denote the index sets
of the subchannels assigned to the cellular network and the
MANET, respectively. Their cardinalities are represented by
K := |A| and

K := |

A|, where K +

K = M.
Spectrum overlay requires initialization where base stations
broadcast to ad hoc nodes the indices of the available sub-
channels and the allowable transmitting node density. This
density can be controlled by distributed adjustments of nodes
transmission probability [7]. In practice, K and

K can be
adapted to the time-varying uplink trafc load to increase the
spectrum-usage efciency.
2) Spectrum Underlay: For spectrum underlay, both coex-
isting networks use all M subchannels. Therefore the trans-
mitters accessing the same subchannel comprises both mobiles
and ad hoc nodes. Consequently, spectrum overlaid networks
are coupled rather than decoupled as for the case of spectrum
overlay.
D. Successive Interference Cancelation
SIC decodes multiuser signals and subtracts them from the
aggregate received signal sequentially [30]. This technique is
effective for mitigating interference in the MANET given the
disparity in received power of multiuser signals due to random
node positions as well as path loss and fading [24]. The SIC
decoding of a users signal treats the signals of undecoded
users as noise. Therefore, for users with equal data rates, it is
optimal to decode multiuser signals in the descending order
of their power [31]. The above decoding order implies that
canceled interferers signals are stronger than the desired one.
This motivates the current simplied SIC model based on the
following assumption.
Assumption 3. The interference power of each interferer
targeted for cancelation must exceed a threshold equal to the
desired signal power multiplied by a factor > 1.
This assumption is made for mathematical tractability. More
accurate SIC models must account for cancelation errors and
the decoding order [26], [30], [32]. Increasing decreases the
average number of canceled interferers and vice versa. This
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
HUANG et al.: SPECTRUM SHARING BETWEEN CELLULAR AND MOBILE AD HOC NETWORKS: TRANSMISSION-CAPACITY TRADEOFF 1259
model is very similar to that in [24] where only path-loss is
considered and interferers within a xed distance are canceled.
III. BACKGROUND
In this section, the shot noise process is introduced for
modeling interference in the coexisting networks. Existing
results on the shot-noise distribution are discussed, which are
useful for analysis in the sequel. Finally, the transmission
capacity and capacity region are dened.
A. Shot Noise Process
A general shot noise process refers to a functional resulting
from feeding a memoryless and linear lter with impulses
derived from a stationary PPP [33]. Such processes have been
widely applied in modeling spatial interference in wireless
networks (see e.g. [7], [34]). Consider a wireless network
where transmitters form a stationary PPP with the density

and narrow-band channels comprise path-loss and fading. For

a receiver A
0
located at the origin, the aggregate interference
power I

is a shot noise process given as

I

X\{F
0
}
PQ
X
|X|

(1)
where F
0
denotes the transmitter for A
0
and {Q
X
} are
i.i.d. fading factors. It is important to note that the pro-
cess \{F
0
} is identically distributed as accordingly to
Slyvnyaks theorem [29]. Analyzing the outage probability
for the transmission from F
0
to A
0
requires deriving the
complementary cumulative density function (CCDF) of I

.
Unfortunately, this function has no closed-form expression [7],
[33].
The CCDF of I

can be bounded using the approach in

[6], [7]. This approach separates the interferers in \{F
0
}
into strong and weak interferers. Dene the process of strong
interferers as
S
= {X \{F
0
} | PQ
X
|X|

> t}, where

each interferer alone guaranteers the outage I

> t. It follows
that the process of weaker interferers is
c
S
:= (\{F
0
})\
S
.
Using these denitions
Pr(I

> t) = Pr (
S
= ) Pr (I
c
S
> t) + Pr (
S
= ) (2)
where I
c
S
:=

X
c
S
PQ
X
|X|

represents the interference

power from weak interferers. From (2), a lower bound on the
outage probability is given as
Pr(I

> t) > Pr (
S
= ) = 1 e

S
(3)
where
S
:= E[|
S
|] is the average number of strong interfer-
ers. The upper bound on Pr(I

> t) is obtained by bounding

the term Pr (I
c
S
> t) in (2) using Chebyshevs inequality
Pr (I
c
S
> t)
var(I
c
S
)
(t E[I
c
S
])
2
, t > E[I
c
S
] (4)
where the right-hand-side of the rst inequality is Chebychevs
upper bound. Closed-form expressions for
S
, var(I
c
S
) and
E[I
c
S
] can be obtained using Campbells theorem [7], [29].
By substituting the resultant expressions into (3) and (4), the
bounds on Pr(I

> t) are obtained in [7, Theorem 2] and

shown in the following lemma.
Lemma 1. The CCDF of the shot noise process in (1) is
bounded as
P

out
(t,

) < Pr(I

> t) < P
u
out
(t,

)
where P

out
(a, b) = 1 exp
_
a

b
_
, P
u
out
(a, b) = 1
(a, b) exp
_
a

b
_
, and
(a, b) =

2
a

b
_
1

1
a

b
_
2

+
,

1
a

b < 1
0, otherwise
(5)
with :=

E[Q

].
B. Transmission Capacity
We dene two related performance metrics for the coexist-
ing networks. As in [7], the networks are assumed to be in-
terference limited and noise is neglected for simplicity. Hence
the SIR measures the reliability of received data packets. Let
SIR and

SIR represent the SIRs at U
0
and R
0
, respectively.
The correct decoding of received packets requires the SIRs
to exceed a threshold . To satisfy this constraint with high
probability, the following outage constraints are applied for
given 0 < 1
P
out
(,

) := Pr(SIR(,

) < )

P
out
(,

) := Pr(

SIR(,

) < )
(6)
where the functions P
out
and

P
out
map the SIRs to the outage
probabilities. The transmission capacities C of the cellular
network and

C of the MANET are dened as the maximum
transmitter densities under the outage constraints in (6). This
denition differs slightly from that in [6], [7] by a linear factor
(1 ), which has a negligible effect on the analysis.
Next, we dene the capacity region R as the set comprising
all combinations (,

) that satisfy the outage constraints in

(6)
R =
_
(,

) R
2
| P
out
(,

) ,

P
out
(,

)
_
. (7)
As mentioned earlier, transmission capacities specify the
boundary points of R.
IV. NETWORK OUTAGE PROBABILITIES
In this section, the outage probabilities for the coexisting
networks are derived for spectrum overlay and underlay with
and without SIC.
For the notation used in this section, the superscripts (a)-
(d) identify four cases: (a) spectrum overlay, (b) spectrum
underlay, (c) spectrum overlay with SIC and (d) spectrum
underlay with SIC.
A. Outage Probabilities: Spectrum Overlay
For spectrum overlay, the SIRs for the coexisting networks
are obtained as follows. Let
(a)

denote the process of

mobile transmitters accessing the th subchannel. Similarly,

(a)
m
represents the process of ad hoc nodes transmitting over
the mth subchannel. Since spectrum overlay decouples the
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
1260 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 7, SEPTEMBER 2009
networks, mobile receivers are interfered with only by unin-
tended mobile transmitters, and ad hoc receivers by unintended
ad hoc transmitters. Without loss of generality, consider the
time slot where the receivers B
0
and R
0
decode signals
from the th and the mth subchannels, respectively, where
A and m

A. The interference power for B
0
can be
written as I
(a)

:= P

X
(a)

G
X
R

X
and that for R
0
as

I
(a)
m
:=

P

X
e

(a)
m
G
X

R

X
. It follows that the SIRs are
given as
Celluar : SIR
(a)

= PWD

/I
(a)

MANET :

SIR
(a)
m
=

P

W

d

I
(a)
m
.
(8)
Given frequency-hopping spread spectrum, multiple sub-
channels assigned to each network contributes a processing
gain for reducing the density of transmitters accessing the
same subchannel. Specically, the densities of
(a)

and

(a)
m
are shown to be inversely proportional to K and

K, respec-
tively, as follows. Dene the mark M
X
of a mobile transmitter
X as the index of the subchannel X accesses in the current
time slot. It follows from the marking theorem [35] that
(a)

is a stationary PPP with the density Pr(M

X
= ) = /K,
which is a thinned process generated by
(a)
. Similarly,

(a)
m
is a stationary PPP with the density

/

K.
Using the above results and (8), the bounds on the out-
age probabilities for spectrum overlay are obtained using
Lemma 1. Note that the distribution of SIR
(a)

is independent
of . Thus the outage probability for the cellular network is
P
(a)
out
= Pr
_
SIR
(a)

<
_
= E
_
Pr
_
I
()

> WD

1
| W, D
__
. (9)
Recognize that I
(a)

is a shot noise process (see Section III-A).

Thus from Lemma 1 and (9), P
(a)
out
is bounded as shown in
the following proposition. Similarly, we obtain the bounds
on the outage probability

P
(a)
out
for the MANET as given in
Proposition 1.
Proposition 1. For spectrum overlay, the outage probabilities
are bounded as
Cellular : E
_
P
l
out
_
WD

,

K
__
P
(a)
out
(K, )
E
_
P
u
out
_
WD

,

K
__
MANET : E
_
P
l
out
_

W

d

K
__

P
(a)
out
(

K,

) E
_
P
u
out
_

W

d

K
__
where P

out
(, ) and P
u
out
(, ) are given in Lemma 1.
Recall from Section II-B that W and

W follow the same
distributions as the fading factors of data links in the cel-
lular network and the MANET, respectively; G is identically
distributed as the fading factors of the interference channels
in the coexisting networks.
Finally, the CDF of D is obtained. Note that the event D t
is equivalent to that there exists at least one base station within
a distance of t from U
0
. Lemma 2 follows.
Lemma 2. The CDF of D is
Pr(D t) = 1 e

b
t
2
. (10)
From (10), the CDF of D depends on the density of base
station
b
. Intuitively, increasing
b
reduces the cell sizes and
thus D and vice versa.
B. Outage Probabilities: Spectrum Underlay
The SIRs for spectrum underlay are obtained as follows.
Dene the transmitter processes
(b)

and

(b)
m
similarly as

(a)

and

(a)
m
in the preceding section. Again, using the mark-
ing theorem,
(b)

and

(b)
m
are shown to be stationary PPPs
with the densities /M and

/M, respectively. For spectrum
underlay, B
0
accessing the th subchannel is interfered with
by two processes of transmitters, namely
(b)

\{U
0
} and

(b)

.
Thus the interference power for B
0
is given as
I
(b)

:= P

X
(b)

\{U
0
}
G
X
R

X
+

P

X
e

(b)

G
X
R

X
. (11)
Similarly, we can write the interference power for R
0
access-
ing the mth subchannel as

I
(b)
m
:= P

X
(b)
m
G
X

R

X
+

P

X
e

(b)

\{T
0
}
G
X

R

X
. (12)
Given the above denitions, the SIRs for spectrum underlay
are readily written as
Celluar : SIR
(b)

= PWD

/I
(b)

MANET :

SIR
(b)
m
=

P

W

d

I
(b)
m
(13)
where , m {1, 2, , M}.
We consider the combined process

:=
(b)

(b)

. A
point in

is either a mobile or ad hoc transmitter. As a result,

its transmission power is a random variable with the support
set {P,

P}. To analyze the shot noise process generated by

,
we derive the distributions of

and the transmission power

of the points in

. These results are summarized below.

Lemma 3.

is a stationary PPP with the density (+

)/M.
Given X

, the transmission power P

X
of X has the
following distribution function
P
X
=

P, w.p.

+

P, w.p.

+

.
(14)
Proof: See Appendix A.
In terms of

, (11) and (12) can be simplied as

I
(b)

:=

X

\{U
0
}
P
X
G
X
R

I
(b)
m
:=

X
m
\{T
0
}
P
X
G
X

R

X
.
(15)
Using Lemma 3 and (15) and following the approach in
Section III-A, the bounds on the outage probabilities are
obtained and shown below.
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
HUANG et al.: SPECTRUM SHARING BETWEEN CELLULAR AND MOBILE AD HOC NETWORKS: TRANSMISSION-CAPACITY TRADEOFF 1261
Proposition 2. For spectrum underlay, the outage probabili-
ties are bounded as follows.
Cellular : E
_
P
l
out
_
WD

,
+

M
__
P
out
(,

)
E
_
P
u
out
_
WD

,
+

M
__
MANET : E
_
P
l
out
_

W

d

M
__


P
out
(,

)
E
_
P
u
out
_

W

d

M
__
where the power ratio := / , and P
l
out
(, ) and P
u
out
(, )
are dened in Lemma 1.
Proof: See Appendix B.
Proposition 2 shows that the outage probability for each
network depends on the transmitter densities of both net-
works. This coupling is due to spectrum underlay and the
resultant mutual interference between the coexisting networks.
As shown in Section V, such coupling may result in smaller
transmission capacities for spectrum underlay than those for
spectrum overlay. Moreover, Proposition 2 also shows that
the outage probabilities for spectrum underlay depend on the
transmission power ratio . The effect of is characterized in
Section V.
C. Outage Probabilities: Spectrum Overlay with SIC
As in the preceding sections, consider B
0
and R
0
accessing
the th and the mth subchannels, respectively. SIC effectively
remove the strongest interferers for B
0
and R
0
. Based on
the SIC model in Section II-D, the interferer process for B
0
conditioned on the link power WD

is

(c)

(WD

) :=
_
X
(a)

\{U
0
} | G
X
R

X
WD

_
(16)
where
(a)

is dened for spectrum overlay in Section IV-A.

Similarly, the conditional interferer process for R
0
is given by

(c)
m
(

W

d

) :=
_
X

(a)

\{T
0
} | G
X

R

W

d

_
(17)
where

(a)

follows from Section IV-A. Thus the conditional

interference power for B
0
and R
0
can be written as
I
(c)

(WD

) := P

X
(c)

(WD

)
G
X
R

I
(c)
m
(

W

d

) :=

P

X
e

(c)
m
(
f
W

d

)
G
X

R

X
.
(18)
It follows that SIRs for spectrum overlay with SIC are
Celluar : SIR
(c)

= PWD

/I
(c)

(WD

)
MANET :

SIR
(c)
m
=

P

W

d

I
(c)
m
(

W

d

).
(19)
The above SIR derivation shows that spectrum overlay with
and without SIC are closely related. This relation is specied
in the following proposition.
Proposition 3. For spectrum overlay with SIC, the bounds on
the outage probabilities P
(c)
out
and

P
(c)
out
can be modied from
those in Proposition 1 by replacing the functions P
l
out
and
P
u
out
with

P
l
out
and

P
u
out
correspondingly, which are dened as

P
l
out
(a, b) := 1 exp
_
a

b
_
(20)

P
u
out
(a, b) := 1 (a, b) exp
_
a

b
_
(21)
where := 1

.
Proof: See Appendix C.
Note that (20) and (21) differ from their counterparts in
Lemma 1 only by the factor . The factor < 1 represents the
SIC advantage of reducing outage probabilities with respect
to the case of no SIC ( = 1). Moreover, decreasing the SIC
factor reduces and thus outage probabilities. Nevertheless,
being too small may invalidate the assumption of perfect
SIC. Specically, small implies small SIR for the process
of decoding interference prior to its cancelation and potentially
results in signicant residual interference after SIC [26].
D. Outage Probabilities: Spectrum Underlay with SIC
The outage probabilities for spectrum underlay with SIC can
be obtained from those for spectrum underlay in Section IV-B.
The procedure is similar to that in the preceding section.
Specically, the conditional interferer processes for B
0
and
R
0
are dened similarly as in (16) and (17)

(d)

(WD

) :=
_
X

\{U
0
} | P
X
G
X
R

X
PWD

(d)
m
(

W

d

) :=
_
X
m
\{T
0
} | P
X
G
X

R

W

d

_
where

is the combined PPP dened Section IV-B, and P

X
is distributed as in Lemma 3. Thus the SIRs can be written as
Celluar : SIR
(d)

= PWD

/I
(d)

(WD

)
MANET :

SIR
(d)
m
=

P

W

d

I
(d)
m
(

W

d

).
(22)
where the conditional interference power
I
(d)

(a) := P

X
(c)

(a)
G
X
R

I
(d)
m
(a) :=

P

X
e

(c)
m
(a)
G
X

R

X
.
(23)
The similar results as in Proposition 3 are obtained for
spectrum underlay with SIC and shown in the following
proposition. Its proof is similar to that for Proposition 3 and
thus omitted.
Proposition 4. For spectrum underlay with SIC, the bounds
on the outage probabilities P
(d)
out
and

P
(d)
out
can be modied
from those in Proposition 2 by replacing the functions P
l
out
and P
u
out
with

P
l
out
and

P
u
out
dened in Proposition 3.
The remarks on Proposition 3 are also valid for Proposition 4.
V. NETWORK CAPACITY TRADEOFF: ASYMPTOTIC
ANALYSIS
Using the results obtained in the preceding section, the
trade-off between the transmission capacities C and

C of the
coexisting networks is characterized in the following theorem
for small target outage probability 0.
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
1262 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 7, SEPTEMBER 2009
Theorem 1. For 0, transmission capacities of the
coexisting networks satisfy

C +C =
M

+O
_

2
_
(24)
where the weights and are given as
4
_

o
= E[

d
2
,
o
= 2E[W

](
b
)
1

u
=
o
(

o
),
u
= (

o
)
o
,
(25)
and depends on if SIC is used
_
= 1, no SIC
1

2
2

, SIC.
(26)
Proof: See Appendix D.
It can be observed that for spectrum overlay the numbers
of subchannels assigned to the coexisting networks, namely
K and

K, do not appear in Theorem 1. The reason is that K
and

K are merged into M based on the equality M = K+

K
(see Appendix D). Theorem 1 shows that the trade-off between
C and

C follows a linear equation. Specically, the slope at
which

C increases with decreasing C is / , which depends
on different network parameters as observed from (25). The
results in Theorem 1 are interpreted using several corollaries
in the sequel.
To facilitate discussion, dene an outage limited network as
one whose transmission capacity is achieved with the outage
constraint being active. For instance, the cellular network is
outage limited if P
out
(C) = . For spectrum overlay, both
the coexisting networks are outage limited. Nevertheless, for
spectrum underlay, it is likely that only one of the two
networks is outage limited as explained shortly. As implied by
the proof for Theorem 1, for spectrum underlay, both networks
are outage limited only if
u
=
u
, where
u
and
u
are given
in (25). Otherwise,
u
>
u
correspond to only the cellular
network being outage limited;
u
<
u
indicates that only the
MANET is outage limited.
Spectrum overlay is shown to be more efcient than spec-
trum underlay as follows. By denitions, the capacity region
in (7) is the region enclosed by the capacity trade-off curve
in (24) and the positive axes of the C-

C coordinates. This
region contains all feasible combinations of the densities
of the coexisting networks. Thus, the size of the capacity
region measures the spectrum-sharing efciency. The capacity
regions for spectrum overlay and underlay are compared in
the following corollary.
Corollary 1. For 0, the capacity region for spectrum
underlay is no larger than that for spectrum overlay. They are
identical if and only if the transmission-power ratio is
= (
o
/
o
)
1

(27)
where
o
and
o
are given in (25).
Proof: See Appendix E.
Corollary 1 shows that spectrum overlay is potentially more
efcient than spectrum underlay due to network coupling
for the latter. Specically, the possibility that a network is
4
The subscripts o and u identify spectrum overlay and underlay, respec-
tively
not outage limited compromises the efciency of spectrum
underlay, which, however, can be compensated by setting
as given in (27). This optimal value of ensures that both
networks are outage limited for the case of spectrum underlay.
The next corollary species the effects of several parameters
on transmission capacities of the coexisting networks.
Corollary 2. For 0, transmission capacities vary with
network parameters as follows.
1) Spectrum overlay: C increases linearly with the base
station density
b
;

C increases inversely with the ad hoc
transmitter-receiver distance

d.
2) Spectrum underlay: If the cellular network is outage
limited, both C and

C increase linearly with the base
station density
b
. Otherwise, both C and

C increase
inversely with the ad hoc transmitter-receiver distance

d.
3) For both spectrum sharing methods, C and

C increase
linearly with and the number of subchannels M, and
inversely with related to SIC.
Finally, we analyze the transmission-capacity gains due to
spatial diversity gains contributed by multi-antennas [36]. To
obtain concrete results, the fading factors W and

W are as-
sumed to follow the chi-squared distributions with the complex
degrees of freedomL and

L, respectively, which are called the
diversity gains [30]. These fading distributions can result from
using spatial diversity techniques such as beamforming over
multi-antenna i.i.d. Rayleigh fading channels [37]. Thus
E[W

] =
(L )
(L)
, E[

] =
(

L )
(

L)
. (28)
The following corollary is obtained by combining Theorem 1,
(28) and the following Kershaws inequalities [38]
_
x +
s
2
_
1s
<
(x +s)
(x + 1)
<
_
x
1
2
+
_
s +
1
4
_1
2
_
1s
where x 1 and 0 < s < 1.
Corollary 3 (Spatial diversity gain). Consider the diversity
gains per link of L and

L for the coexisting cellular and ad
hoc networks, respectively.
1) Spectrum overlay: The spatial diversity gains multiply
C by a factor between (L 1)

and L

, and

C by a
factor between (

L 1)

and

L

.
2) Spectrum underlay: The spatial diversity gains multi-
ply both C and

C by a factor between (L1)

and L

if the cellular network is outage limited, or otherwise

between (

L 1)

and

L

.
Note that similar results are obtained in [22] for a stand-
alone MANET by using a more complicated method than the
current one based on Kershaws inequalities.
VI. SIMULATION AND NUMERICAL RESULTS
In this section, the tightness of the bounds on outage
probabilities derived in Section IV is evaluated using simu-
lation. Moreover, the asymptotic transmission capacity trade-
off curves obtained in Theorem 1 are compared with the
non-asymptotic ones generated by simulation. The simulation
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
HUANG et al.: SPECTRUM SHARING BETWEEN CELLULAR AND MOBILE AD HOC NETWORKS: TRANSMISSION-CAPACITY TRADEOFF 1263
10
5
10
4
10
3
10
2
10
3
10
2
10
1
10
0
User Density per SubChannel
O
u
t
a
g
e

P
r
o
b
a
b
i
l
i
t
y

o
f

C
e
l
l
u
l
a
r

N
e
t
w
o
r
k


Simulation (w/o SIC)
Lower Bound (w/o SIC)
Upper Bound (w/o SIC)
Simulation (w/ SIC)
Lower Bound (w/ SIC)
Upper Bound (w/ SIC)
(a) Spectrum overlay: cellular network
10
5
10
4
10
3
10
2
10
3
10
2
10
1
10
0
Node Density per SubChannel
O
u
t
a
g
e

P
r
o
b
a
b
i
l
i
t
y

o
f

A
d

H
o
c

N
e
t
w
o
r
k


Simulation (w/o SIC)
Lower Bound (w/o SIC)
Upper Bound (w/o SIC)
Simulation (w/ SIC)
Lower Bound (w/ SIC)
Upper Bound (w/ SIC)
(b) Spectrum overlay: ad hoc Network
10
5
10
4
10
3
10
2
10
3
10
2
10
1
10
0
Transmitter Density per SubChannel
O
u
t
a
g
e

P
r
o
b
a
b
i
l
i
t
y

o
f

C
e
l
l
u
l
a
r

N
e
t
w
o
r
k


Simulation (w/o SIC)
Lower Bound (w/o SIC)
Upper Bound (w/o SIC)
Simulation (w/ SIC)
Lower Bound (w/ SIC)
Upper Bound (w/ SIC)
(c) Spectrum underlay: cellular network
10
5
10
4
10
3
10
2
10
3
10
2
10
1
10
0
Transmitter Density per SubChannel
O
u
t
a
g
e

P
r
o
b
a
b
i
l
i
t
y

o
f

A
d

H
o
c

N
e
t
w
o
r
k


Simulation (w/o SIC)
Lower Bound (w/o SIC)
Upper Bound (w/o SIC)
Simulation (w/ SIC)
Lower Bound (w/ SIC)
Upper Bound (w/ SIC)
(d) Spectrum underlay: ad hoc Network
Fig. 2. Comparison between the theoretical bounds on outage probabilities and the simulated values. For spectrum underlay, the densities of mobile and ad
hoc transmitters are set equal. The sum density is referred to in the gures for spectrum underlay as the transmitter density.
procedure summarized below is similar to that in [39]. The
typical base station (or the ad hoc receiver) of the coexisting
network lies at the centers of two overlapping disks, which
contain interfering transmitters (either ad hoc nodes, users
or both) and base stations respectively. Both the transmitters
and the base stations follow the Poisson distribution with
the mean equal to 200. The disk radiuses are adjusted to
provide the desired densities of transmitters or base stations.
The simulation parameters are set as d = 5 m, = 3 or 4.8
dB, = 4,
b
= 10
3
, = 2 dB, and = 5 dB.
Fig. 2 compares the bounds on outage probabilities in
Section IV and the simulated values. As observed from Fig. 2,
for all cases, the outage probabilities converge to their lower
bounds as the transmitter densities decrease; the upper and
lower bounds differ by approximately constant multiplicative
factors. Fig. 2 also shows that SIC reduces outage probabilities
by a factor of about 0.54 approximately equal to in Propo-
sition 3 and Proposition 4. Moreover, SIC loosens the bounds
on outage probabilities for relatively large transmitter densities
since SIC reduces the number of strong interferers to each
receiver. Finally, outage probabilities become proportional to
transmitter densities as they decrease.
Fig. 3 compares the asymptotic transmission-capacity trade-
off curves in Theorem 1 and those generated by simulations
for the target outage probability = 10
2
. In Fig. 3(b) for
the case of SIC, the bounds on the asymptotic trade-off curves
correspond to those on as given in Theorem 1. By comparing
Fig. 3(a) and Fig. 3(b), the capacity regions for spectrum
overlay are larger than those for spectrum underlay. For the
case of no SIC, the asymptotic results closely match their
simulated counterparts. When SIC is used, the capacity trade-
off curve generated by simulation is close to the corresponding
asymptotic upper bound. In particular, for spectrum overlay
with SIC, the simulation results are practically identical to
their asymptotic upper bounds. In summary, the asymptotic
results derived in Section V are useful for characterizing
network transmission capacities in the non-asymptotic regime.
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
1264 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 7, SEPTEMBER 2009
0 1 2 3 4 5 6 7 8 9
x 10
5
0
1
2
3
4
5
6
x 10
5
Capacity of Ad Hoc Network / M
C
a
p
a
c
i
t
y

o
f

C
e
l
l
u
l
a
r

N
e
t
w
o
r
k

/

M


Without SIC (Simulation)
Without SIC (Asymptotic)
With SIC (Simulation)
With SIC (Asymptotic Lower Bound)
With SIC (Asymptotic Upper Bound)
(a) Spectrum overlay
0 1 2 3 4 5 6 7 8 9
x 10
5
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
x 10
5
Capacity of Ad Hoc Network / M
C
a
p
a
c
i
t
y

o
f

C
e
l
l
u
l
a
r

N
e
t
w
o
r
k

/

M


Without SIC (Simulation)
Without SIC (Asymptotic)
With SIC (Simulation)
With SIC (Asymptotic Lower Bound)
With SIC (Asymptotic Upper Bound)
(b) Spectrum underlay
Fig. 3. Comparison between the asymptotic and the simulated transmission-
capacity trade-off curves for the coexisting networks using (a) spectrum
overlay or (b) spectrum underlay
VII. CONCLUSION
In this paper, the transmission-capacity trade-off between
the coexisting cellular and ad hoc networks is analyzed for
different spectrum sharing methods. To this end, bounds on
outage probabilities for both networks are derived for spec-
trum overlay and underlay with and without SIC. For small
target outage probability, the transmission capacities of the
coexisting networks are shown to satisfy a linear equation,
whose coefcients are derived for the cases considered above.
These results provide a theoretical basis for adapting the
node density of the ad hoc network to the dynamic trafc
in cellular uplink under the outage constraints. The trade-
off relationship suggests that transmission capacities of the
coexisting networks can be increased by adjusting various
parameters such as decreasing the distances between intended
ad hoc transmitters and receivers, increasing the base station
density and link diversity gains, or by employing SIC. In
particular, SIC increases the transmission capacities by a linear
factor that depends on the interference power threshold for
qualifying canceled interferers. Simulation results show that
the derived bounds on outage probabilities are tight and the
asymptotic linear capacity trade-off is valid even in the non-
asymptotic regime.
This paper opens several issues for future research on
spectrum sharing between networks, including the impact
of cognitive radio, the capacity trade-off between competing
networks, and the extension of the current results to more
realistic non-homogeneous network architectures.
APPENDIX
A. Proof for Lemma 3
Since

and

are both stationary PPPs, it follows from

the superposition property of Poisson processes [35] that the
combined process

is also a stationary PPP with the density

(+

)/M, which is equal to the sum of those of

and

.
This proves the rst part of Lemma 3.
Consider X

. Let B(X, r) denote a disk cen-

tered at X and with the radius r, thus B(X, r) =
{Y R
2
||Y X| r }. Denote the area of B(X, r) as
A(B(X, r)). Thus
Pr(P
X
= P | X) = lim
r0
Pr
_

(b)

B(X, r) =
_
Pr (

B(X, r) = )
= lim
r0
1 exp
_

M
A(B(X, r))
_
1 exp
_
+

M
A(B(X, r))
_
= lim
r0
1 e

r
2

M
1 e

r
2
(+

)
M
=

+

.
Therefore Pr(P
X
=

P | X) = 1 Pr(P
X
= P | X) =

.
This completes the proof.
B. Proof for Proposition 2
The following proof focuses on the cellular network. The
proof for the results concerning the MANET is similar and
thus omitted. Based on

, we dene a marked PPP

[35],
where the mark for a point X

is its transmission power

P
X
and the fading factor G
X
(with respect to B
0
)

:=
_
(X, P
X
, G
X
) | X

\{U
0
}, P
X
{, }, G
X
R
+
_
.
Following the approach discussed in Section IV-A,

is
divided into a strong-interferer sub-process conditioned on the
link power WD

, denoted as

S
(w, d) and given as

S
(WD

) :=
_
(X, P
X
, G
X
)

(X, P
X
, G
X
)

,
P
X
|X|

G
X
> PWD

1
_
and the weak-interferer process dened as

c
S
(WD

) =

S
(WD

). Thus, the sum interference power from weak

interferers to the typical base station B
0
can be written
as I
c
S
(WD

) =

(X,P
X
,G
X
)

c
S
(WD

)
P
X
|X|

G
X
,
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
HUANG et al.: SPECTRUM SHARING BETWEEN CELLULAR AND MOBILE AD HOC NETWORKS: TRANSMISSION-CAPACITY TRADEOFF 1265
which is a shot noise processes. To use the results in
Lemma 1 for bounding the outage probability, it is sufcient
to obtain
S
(WD

) := E[|

S
(WD

)|], E[I
c
S
(WD

)]
and var [I
c
S
(WD

)]. Using the marking theorem [35] and

Lemma 3,

S
(a) =
2( +

)
M

Pr (P
X
= P)

_
0
(
g
a
)
1

_
0
rf
G
(g)drdg+
Pr
_
P
X
=

P
_
_

0
_
(
1
a
1
g)
1

0
rf
G
(g)drdg

=
a

( +

)
M
(29)
where f
G
(g) is the probability density function (PDF) of
G
X
and is dened in Lemma 1. Next, E[I
c
S
(w, d)] and
var [I
c
S
(w, d)] are derived using Campbells theorem [35] and
Lemma 3 as shown on the next page. Substituting (29), and
(30) and (31) into (6) and (4) gives the desired results.
C. Proof for Proposition 3
Only the bounds on P
out
are proved. The proof for those
on

P
out
is similar and thus omitted. Dene the process
of strong interferers after SIC as
S
(WD

) := {X

(c)

\{U
0
} |
1
WD

PG
X
|X|

WD

}. Note
that WD

>
1
WD

since > 1 and > 1.

Thus, the process of weak interferers can be dened as

c
S
(WD

) :=
(c)

\
S
(WD

), which is observed to be
identical to the counterpart for spectrum overlay without SIC
considered in Section B. Since
c
S
(WD

)
S
(WD

) =
,
c
S
(WD

) and
S
(WD

) are independent processes

based on a basic property of Poisson processes [35]. From
the discussion in Section III-A, the exponential terms in (3)
and (4) depends only on
S
(WD

), and the function (, )

only on
c
S
(WD

). Since
c
S
(w, d) is invariant to SIC, and

c
S
(w, d) and
S
(w, d) are independent, the bounds on P
out
in
Lemma 1 can be extended to the case of SIC by replacing the
exponential terms in (3) and (4) with exp(E[|
S
(w, d)|]),
where E[|
S
(w, d)|] is obtained using Campbells theorem
E[|
S
(w, d)|] =2
_

0
_
(w
1
d

g)
1

(
1
w
1
d

g)
1

rf
G
(g)drdg
=w

d
2

K
and is dened in the statement of the proposition.
D. Proof for Theorem 1
1) Spectrum Overlay: The convergence 0 implies
0 and

0. Using the series representation of the PDF of a
power shot-noise process [33], the asymptotes of the outage
probabilities follow from [7, Theorem 2]
P
out
=

K
E
_
W

E
_
D
2

+O
_

2
_

P
out
=

K
E
_

_

d
2
+O
_

2
_
. (32)
By using Lemma 2, the term E
_
D
2

in (32) is obtained as
E
_
D
2

=
_

0
2
b
t
3
e

b
t
2
dt
=
_

0

b
te

b
t
dt =
(2)

b
=
2

b
. (33)
Combining (6), (32), and (33) gives the desired asymptotic
capacity trade-off function for spectrum overlay.
2) Spectrum Underlay: By using the series expression of
the PDF of the power shot noise [33] as well as Proposition 1,
P
out
(,

)=

M
E[W

]E[D
2
] +O(max(
2
,

2
))(34)

P
out
(,

)=

M
E[

d
2
+O(max(
2
,

2
)). (35)
For 0, the transmission capacities C and

C satisfy the
constraints P
out
(C/M,

C/M) and

P
out
(C/M,

C/M) .
By combining these constraints, (34) and (35)
C +

C
M
max
_
E[W

]E[D
2
],

E[

d
2
_
= +O(
2
).
The desired result follows from the above equation.
3) Spectrum Sharing with SIC: Consider spectrum overlay
with SIC. By canceling the strongest interferers using SIC, the
PDF upper-tail of the power shot noise process is trimmed
and its series expansion is difcult to nd [33]. Nevertheless,
the asymptotic transmission capacities can be characterized by
expanding the bounds on P
out
in Proposition 3. Specically
P
l
out
(/K) =

K

E[W

]E[D
2
] +O(
2
)
P
u
out
(/K) = 1 E
__
1

2
W

D
2

K
+O(
2
)
_

_

K

D
2
+O(
2
)
__
=
_
2
2

1

_
E[W

]E[D
2
]

K
+O(
2
).
Thus
P
out
(/K) = E[W

]E[D
2
]

K
(36)
where
_
1

_

_
2
2

_
. Similarly

P
out
(

/

K) = E[

d
2

K
. (37)
The desired results for spectrum overlay with SIC are obtained
by combining (6), (36), and (37). The results for spectrum
underlay with SIC are derived following a similar procedure.
E. Proof for Corollary 1
First, the capacity region for spectrum underlay is proved to
be no larger than for spectrum overlay. It is sufcient to prove
that
u

o
and
u

o
, which follow from (25). Next,
substituting (27) into (25) results in
u
=
o
and
u
=
o
.
This proves the second claim in the theorem statement.
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
1266 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 27, NO. 7, SEPTEMBER 2009
E[I
c
S
(a)] =
2( +

)
M

Pr (P
X
= P)

_
0

_
(
g
a
)
1

(Pr

g)rf
G
(g)drdg + Pr
_
P
X
=

P
_

_
0

_
(
g
a
)
1

(

Pr

g)rf
G
(g)drdg

=
P
1
_
+

M
_
(a
1
)
1

1
, (30)
var[I
c
S
(a)] =
2( +

)
M

Pr (P
X
= P)

_
0

_
(
g
a
)
1

(Pr

g)
2
rf
G
(g)drdg + Pr
_
P
X
=

P
_

_
0

_
(
g
a
)
1

(

Pr

g)
2
rf
G
(g)drdg

=
P
2

2
_
+

M
_
(a
1
)
2

2
. (31)
REFERENCES
[1] Federal Communications Commission, Spectrum policy task force,
Nov. 2002. (Rep. ET Docket no. 02-135).
[2] H. Holma and A. Toskala, WCDMA for UMTS: Radio Access for Third
Generation Mobile Communications. John Wiley & Sons,, 2004.
[3] P. Marques, J. Bastos, and A. Gameiro, Opportunistic use of 3G
uplink licensed bands, in Proc. IEEE Intl. Conf. on Communications,
pp. 35883592, May 2008.
[4] D. Kim and D. Jeong, Capacity unbalance between uplink and down-
link in spectrally overlaid narrow-band and wide-band CDMA mobile
systems, IEEE Trans. Veh. Technol., vol. 49, pp. 10861093, July 2000.
[5] I. F. Akyildiz, W. Lee, M. C. Vuran, and S. Mohanty, Next gen-
eration/dynamic spectrum access/cognitive radio wireless networks: a
survey, Computer Networks, vol. 50, no. 13, pp. 21272159, 2006.
[6] S. P. Weber, X. Yang, J. G. Andrews, and G. de Veciana, Transmission
capacity of wireless ad hoc networks with outage constraints, IEEE
Trans. Inform. Theory, vol. 51, pp. 409102, Dec. 2005.
[7] S. P. Weber, J. G. Andrews, and N. Jindal, The effect of fading, channel
inversion, and threshold scheduling on ad hoc networks, IEEE Trans.
Inform. Theory, vol. 53, pp. 41274149, Nov. 2007.
[8] M. K. Simon, J. K. Omura, R. A. Scholtz, and B. K. L. (Eds.), Spread
Spectrum Communications Handbook. Mc-Graw Hill, 1994.
[9] Q. Zhao and B. M. Sadler, A survey of dynamic spectrum access,
IEEE Signal Processing Mag., vol. 24, pp. 7989, May 2007.
[10] R. Menon, R. M. Buehrer, and J. H. Reed, Outage probability based
comparison of underlay and overlay spectrum sharing techniques, in
Proc. IEEE Intrl. Symp. on New Frontiers in Dynamic Spectrum Access
Networks, pp. 101109, Nov. 2005.
[11] Q. Zhao, L. Tong, A. Swami, and Y. Chen, Decentralized cognitive
MAC for opportunistic spectrum access in ad hoc networks: A POMDP
framework, IEEE Journal on Sel. Areas in Communications, vol. 25,
pp. 589600, Apr. 2007.
[12] B. Wild and K. Ramchandran, Detecting primary receivers for cognitive
radio applications, in Proc. First IEEE Interl. Symp. on New Frontiers
in Dynamic Spectrum Access Networks, pp. 124130, Nov. 2005.
[13] I. Howitt, WLAN and WPAN coexistence in UL band, IEEE Trans.
Veh. Technol., vol. 50, pp. 111424, July 2001.
[14] S. Han, S. Lee, S. Lee, and Y. Kim, Outage probability analysis of
WPAN under coexistence environments in fading channels, in Proc.
Intrl. Confer. on Inform. Networking, pp. 14, Jan. 2008.
[15] N. Thanthry, M. S. Ali, R. Bhagavathula, and R. Pendse, WLAN and
ad-hoc network coexistence, in Proc. IEEE Veh. Technol. Conf., vol. 7,
pp. 50315035, Sept. 2004.
[16] N. Golmie, R. E. V. Dyck, A. Soltanian, A. Tonnerre, and O. Rbala, In-
terference evaluation of bluetooth and IEEE 802.11b systems, Wireless
Networks, vol. 9, pp. 201211, May 2003.
[17] L. Angrisani, M. Bertocco, D. Fortin, and A. Sona, Experimental study
of coexistence issues between IEEE 802.11b and IEEE 802.15.4 wireless
networks, IEEE Trans. Instrum. Meas., vol. 57, pp. 15141523, Aug.
2008.
[18] R. Etkin, A. Parekh, and D. Tse, Spectrum sharing for unlicensed
bands, IEEE J. Select. Areas Commun., vol. 25, pp. 517528, Apr.
2007.
[19] V. Chandrasekhar and J. G. Andrews, Uplink capacity and interference
avoidance for two-tier femtocell networks, to appear in IEEE Trans.
Wireless Commun..
[20] C. Yin, L. Gao, and S. Cui, Scaling laws of overlaid wireless networks:
A cognitive radio network vs. a primary network, submitted to IEEE
Trans. Networking, May 2008.
[21] P. Gupta and P. R. Kumar, The capacity of wireless networks, IEEE
Trans. Inform. Theory, vol. 46, pp. 388404, Mar. 2000.
[22] A. M. Hunter, J. G. Andrews, and S. P. Weber, Transmission capacity of
ad hoc networks with spatial diversity, IEEE Trans. Wireless Commun.,
vol. 7, no. 12, Part 1, pp. 50585071, Dec. 2008.
[23] N. Jindal, J. G. Andrews, and S. P. Weber, Bandwidth partitioning in
decentralized wireless networks, IEEE Trans. Wireless Commun. vol.
7, no. 12, Part 2, pp. 54085419, Dec. 2008.
[24] S. P. Weber, J. G. Andrews, X. Yang, and G. de Veciana, Transmission
capacity of wireless ad hoc networks with successive interference
cancelation, IEEE Trans. Inform. Theory, vol. 53, pp. 27992814, Aug.
2007.
[25] K. Huang, J. G. Andrews, R. W. Heath, Jr., D. Guo, and R. A.
Berry, Spatial interference cancelation for mobile ad hoc networks,
submitted to IEEE Trans. Inform. Theory, May 2008. (ArXiv:
http://arxiv.org/abs/0807.1773).
[26] J. G. Andrews, Interference cancellation for cellular systems: a con-
temporary overview, IEEE Wireless Commun., vol. 12, pp. 1929, Apr.
2005.
[27] F. Baccelli, B. Blaszczyszyn, and P. Muhlethaler, An ALOHA protocol
for multihop mobile wireless networks, IEEE Trans. Inform. Theory,
vol. 52, pp. 42136, Feb. 2006.
[28] F. Baccelli, Stochastic geometry: a tool for modeling telecommunica-
tion networks (INRIA-ENS webpage), http://www.di.ens.fr/.mistral/sg/.
[29] D. Stoyan, W. S. Kendall, and J. Mecke, Stochastic Gemoetry and its
Applications. Wiley, 2nd ed., 1995.
[30] D. Tse and P. Viswanath, Fundamentals of Wireless Communication.
Cambridge University Press, 2005.
[31] D. Warrier and U. Madhow, On the capacity of cellular CDMA with
successive decoding and controlled power disparities, in Proc. IEEE
Veh. Technol. Conf., vol. 3, pp. 18731877, May 1998.
[32] T. M. Cover and J. A. Thomas, Elements of Information Theory. John
Wiley, 1991.
[33] S. B. Lowen and M. C. Teich, Power-law shot noise, IEEE Trans.
Inform. Theory, vol. 36, pp. 130218, Nov. 1990.
[34] E. S. Sousa, Interference modeling in a direct-sequence spread-
spectrum packet radio network, IEEE Trans. Commun., vol. 38,
pp. 147582, Sept. 1990.
[35] J. F. C. Kingman, Poisson processes. Oxford University Press, 1993.
[36] A. Goldsmith, Wireless Communications. Cambridge University Press,
2005.
[37] A. Paulraj, R. Nabar, and D. Gore, Introduction to Space-Time Wireless
Communications. Cambridge University Press, 2003.
[38] D. Kershaw, Some extensions of W. Gautschis inequalities for the
Gamma function, Math. of Computation, vol. 41, pp. 607611, Oct.
1983.
[39] S. P. Weber and M. Kam, Computational complexity of outage prob-
ability simulations in mobile ad-hoc networks, in Proc. Conf. Inform.
Sciences and Systems, Mar. 2005.
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.
HUANG et al.: SPECTRUM SHARING BETWEEN CELLULAR AND MOBILE AD HOC NETWORKS: TRANSMISSION-CAPACITY TRADEOFF 1267
Kaibin Huang (S05M08) received the B.Eng.
(rst-class hons.) and the M.Eng. from the National
University of Singapore in 1998 and 2000, respec-
tively, and the Ph.D. degree from The University of
Texas at Austin (UT Austin) in 2008, all in electrical
engineering.
Since Mar. 2009, he has been an assistant pro-
fessor in the School of Electrical and Electronic
Engineering at Yonsei University, Seoul, Korea.
From Jun. 2008 to Feb. 2009, he was a Postdoctoral
Research Fellow in the Department of Electrical and
Computer Engineering at the Hong Kong University of Science and Technol-
ogy. From Nov. 1999 to Jul. 2004, he was an Associate Scientist at the Institute
for Infocomm Research in Singapore. His research interests focus on limited
feedback techniques for wireless networks. Dr. Huang received the Motorola
Partnerships in Research Grant, the University Continuing Fellowship at UT
Austin, and the Best Student Paper award at IEEE GLOBECOM 2006.
Vincent K. N. Lau obtained B.Eng (Distinction 1st
Hons) from the University of Hong Kong (1989-
1992) and Ph.D. from Cambridge University (1995-
1997). He was with HK Telecom (PCCW) as system
engineer from 1992-1995 and Bell Labs - Lucent
Technologies as member of technical staff from
1997-2003. He then joined the Department of ECE,
Hong Kong University of Science and Technol-
ogy (HKUST) as Associate Professor. His current
research interests include the robust and delay-
sensitive cross-layer scheduling of MIMO/OFDM
wireless systems with imperfect channel state information, cooperative and
cognitive communications, dynamic spectrum access as well as stochastic
approximation and Markov Decision Process.
Yan Chen received her B.Sc degree from Chu
Kochen Honored College, Zhejiang University,
Hangzhou, China, in 2004. She is expected to re-
ceive her Ph.D degree in information and com-
munication engineering from Zhejiang University
in 2009. Since Jan 2008, She has been a visiting
researcher in the group of Prof. Vincent Lau in
HKUST. Her current research interests lie in com-
bined information theory and queueing theory in
wireless communications, with particular emphasis
on exploiting communication opportunities via co-
operation and cognition.
Authorized licensed use limited to: Yonsei University. Downloaded on September 21, 2009 at 05:45 from IEEE Xplore. Restrictions apply.